Novelty Exploration and Adaptation with Reinforcement Learning agents feat. human

How should an AI adapt to changes by detecting novelties, learning from them, and seek for help from human when needed?

Projects

Collaboration with GTRI.

Paper Accepted by RLC 2024 Workshop.

Safe Online Learning and Adaptation

How should an AI learn how to find "Point of Interest" in a changing environment so that it can learn to actively explore them and best improve its knowledge and adapt to the changes?

Jonathan Balloch, Zhiyu Lin, Robert Wright, Xiangyu Peng, Mustafa Hussain, Aarun Srinivas, Julia Kim, Mark Riedl

Available on Arxiv

Neuro-Symbolic World Models for Adapting to Open World Novelty

A symbolic world model that helps neural model to learn better by imagining the future!

Jonathan Balloch, Zhiyu Lin, Mustafa Hussain, Aarun Srinivas, Robert Wright, Xiangyu Peng, Julia Kim, Mark Riedl

Oral at AAAI 2022 Spring Symposium Series

Novgrid: A flexible grid world for evaluating agent response to novelty

Novgrid makes studying novelties modular and streamlined!

Zhiyu Lin, Brent Harrison, Aaron Keech, Mark O Riedl

Available on Arxiv

Explore, Exploit or Listen: Combining Human Feedback and Policy Model to Speed up Deep Reinforcement Learning in 3D Worlds

This Reinforcement Learning agent knows both how to learn from rewards and seek for human feedback - only when needed!